2026-06-23T02:34:29.955Z http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai

ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai:1:11986 2026-06-14T22:58:44Z 1

The CO Pollution Prediction with Recurrent Neural Networks Optimized by Modified Firefly Algorithm 2026 http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai/record/1/11986 https://doi.org/10.1007/978-3-032-22059-2_8 N. Bačanin DŽakula B. Radomirović V. Željković M. Antonijević M. Živković T. Živković V. Gajić M. Jo V. Simić This paper addresses the prediction of carbon monoxide (CO) pollution using recurrent neural networks optimized by a modified firefly algorithm. Since CO concentration can change rapidly under the influence of traffic, meteorological conditions, and industrial activity, accurate forecasting is important for air quality monitoring and environmental protection. The proposed approach uses recurrent neural networks for time-series prediction, while the modified firefly algorithm is applied to improve hyperparameter tuning and model performance. The model was tested on real air-quality monitoring data and showed promising results for estimating atmospheric CO levels. conferenceObject 99 112 10.1007/978-3-032-22059-2_8 Soft Computing and Its Engineering Applications, Communications in Computer and Information Science, vol. 2873